Action-value function

upshard

Q^{\pi}(s, a) = \mathbb{E}_{\pi}\!\left[G_t \mid S_t = s, A_t = a\right]

How good one action is in one state, assuming you behave normally afterwards. Unlike the state value it names a choice, so a greedy policy can be read straight off it without a model of the environment.

Symbols

$Q^{\pi}$: The action-value function under policy pi.
$a$: The action taken in state s.

Save your progress

Your account

Delete your account?

Action-value function

Keyboard shortcuts